Context Store Skill

What & When

是什么与何时使用

What: Commands for uploading, downloading, querying and deleting documents with metadata and tags through the Context Store server. Also linking to one another via relations.

When to use:

Store important documents for retrieval in future sessions
Build a knowledge base of project documentation
Share architecture, API specs, or design documents across different work sessions
Query documents by tags to find relevant information

Key Benefits:

Cross-session persistence
Tag-based organization with AND logic
Simple JSON output

是什么：通过Context Store服务器进行文档上传、下载、查询、删除（附带元数据和标签）以及建立文档间关联的命令集。

适用场景：

存储重要文档以便在未来会话中检索
构建项目文档知识库
在不同工作会话间共享架构文档、API规范或设计文档
通过标签查询文档以找到相关信息

核心优势：

跨会话持久化存储
基于AND逻辑的标签分类
简洁的JSON输出

Quick Reference

快速参考

CRITICAL: Always use absolute paths - NEVER use

cd

:

重要提示：请始终使用绝对路径——切勿使用

cd

命令：

doc-push

- Upload Documents

doc-push

- 上传文档

bash

uv run --script <skill-root>/commands/doc-push <file> [--tags TEXT] [--description TEXT]

bash

uv run --script <skill-root>/commands/doc-push <file> [--tags TEXT] [--description TEXT]

Example: uv run --script <skill-root>/commands/doc-push specs.md --tags "api,v2"

**Use when**: Store a document for future reference.

**适用场景**：存储文档以供后续查阅。

doc-query

- Search by Tags/Name

doc-query

- 按标签/名称搜索

bash

uv run --script <skill-root>/commands/doc-query [--tags TEXT] [--name TEXT] [--include-relations]

bash

uv run --script <skill-root>/commands/doc-query [--tags TEXT] [--name TEXT] [--include-relations]

Example: uv run --script <skill-root>/commands/doc-query --tags "api,v2"

With relations: uv run --script <skill-root>/commands/doc-query --tags "api" --include-relations

**Use when**: Find documents by tags (AND logic) or name patterns.
**Option**: `--include-relations` adds relations to each document in response.

**适用场景**：通过标签（AND逻辑）或名称模式查找文档。
**可选参数**：`--include-relations`会在结果中添加文档间的关联信息。

doc-search

- Semantic Search

doc-search

- 语义搜索

bash

uv run --script <skill-root>/commands/doc-search "<query>" [--limit INT] [--include-relations]

bash

uv run --script <skill-root>/commands/doc-search "<query>" [--limit INT] [--include-relations]

Example: uv run --script <skill-root>/commands/doc-search "how to configure authentication"

With relations: uv run --script <skill-root>/commands/doc-search "auth" --include-relations

**Use when**: Find documents by meaning using natural language queries.
**Note**: Requires semantic search enabled on server. Returns section offsets for partial reads.
**Option**: `--include-relations` adds relations to each result.

**适用场景**：通过自然语言查询按文档语义查找相关内容。
**注意**：需要服务器启用语义搜索功能。返回结果包含用于部分读取的段落偏移量。
**可选参数**：`--include-relations`会在结果中添加文档间的关联信息。

doc-info

- Get Document Metadata & Relations

doc-info

- 获取文档元数据与关联信息

bash

uv run --script <skill-root>/commands/doc-info <document-id>

bash

uv run --script <skill-root>/commands/doc-info <document-id>

Example: uv run --script <skill-root>/commands/doc-info doc_abc123

**Use when**: View metadata and relations for a specific document without downloading it.
**Output**: Includes `relations` field with parent/child/related document links.

**适用场景**：无需下载文档即可查看特定文档的元数据和关联信息。
**输出**：包含`relations`字段，展示父文档/子文档/相关文档的链接。

doc-read

- Read Text Documents

doc-read

- 读取文本文档

bash

uv run --script <skill-root>/commands/doc-read <document-id> [--offset INT] [--limit INT]

bash

uv run --script <skill-root>/commands/doc-read <document-id> [--offset INT] [--limit INT]

Example: uv run --script <skill-root>/commands/doc-read doc_abc123

Partial: uv run --script <skill-root>/commands/doc-read doc_abc123 --offset 2000 --limit 1000

**Use when**: Output text document content directly to stdout (text files only).
**Partial reads**: Use `--offset` and `--limit` to retrieve specific sections (useful with semantic search results).

**适用场景**：将文本文档内容直接输出到标准输出（仅支持文本文件）。
**部分读取**：使用`--offset`和`--limit`参数检索特定段落（配合语义搜索结果使用非常实用）。

doc-pull

- Download Documents

doc-pull

- 下载文档

bash

uv run --script <skill-root>/commands/doc-pull <document-id> [-o PATH]

bash

uv run --script <skill-root>/commands/doc-pull <document-id> [-o PATH]

Example: uv run --script <skill-root>/commands/doc-pull doc_abc123 -o specs.md

**Use when**: Retrieve a document by its ID.

**适用场景**：通过文档ID检索并下载文档。

doc-delete

- Remove Documents

doc-delete

- 删除文档

bash

uv run --script <skill-root>/commands/doc-delete <document-id>

bash

uv run --script <skill-root>/commands/doc-delete <document-id>

Example: uv run --script <skill-root>/commands/doc-delete doc_abc123

**Use when**: Permanently remove a document.

**适用场景**：永久删除文档。

doc-link

- Manage Document Relations

doc-link

- 管理文档关联

bash

uv run --script <skill-root>/commands/doc-link --types                     # List relation types
uv run --script <skill-root>/commands/doc-link --create <from> <to> [opts] # Create relation
uv run --script <skill-root>/commands/doc-link --update <id> --note "..."  # Update note
uv run --script <skill-root>/commands/doc-link --remove <id>               # Remove relation

Use when: Link documents together in parent-child or peer relationships. Options for --create:

```
--type
```
- Required:
```
parent-child
```
(hierarchical) or
```
related
```
(peer)
```
--from-to-note
```
- Note on edge from source to target (source's note about target)
```
--to-from-note
```
- Note on edge from target to source (target's note about source)

bash

uv run --script <skill-root>/commands/doc-link --types                     # List relation types
uv run --script <skill-root>/commands/doc-link --create <from> <to> [opts] # Create relation
uv run --script <skill-root>/commands/doc-link --update <id> --note "..."  # Update note
uv run --script <skill-root>/commands/doc-link --remove <id>               # Remove relation

适用场景：建立文档间的父子关系或同级关系。 --create参数选项：

```
--type
```
- 必填：
```
parent-child
```
（层级归属）或
```
related
```
（同级关联）
```
--from-to-note
```
- 源文档到目标文档的关联备注（源文档对目标文档的说明）
```
--to-from-note
```
- 目标文档到源文档的关联备注（目标文档对源文档的说明）

Typical Workflows

典型工作流

Store and Retrieve

存储与检索

bash

undefined

bash

undefined

Upload with tags

uv run --script <skill-root>/commands/doc-push specs.md --tags "api,v2"

Find it later

uv run --script <skill-root>/commands/doc-query --tags "api,v2"

Download it

uv run --script <skill-root>/commands/doc-pull doc_abc123

undefined

uv run --script <skill-root>/commands/doc-pull doc_abc123

undefined

Build Knowledge Base

构建知识库

bash

undefined

bash

undefined

Upload multiple documents with consistent tags

uv run --script <skill-root>/commands/doc-push architecture.md --tags "design,mvp" uv run --script <skill-root>/commands/doc-push api-spec.md --tags "api,mvp"

Query by project phase

uv run --script <skill-root>/commands/doc-query --tags "mvp"

undefined

uv run --script <skill-root>/commands/doc-query --tags "mvp"

undefined

Semantic Search + Partial Read

语义搜索+部分读取

bash

undefined

bash

undefined

Search by meaning

uv run --script <skill-root>/commands/doc-search "how to authenticate users"

Returns: {"results": [{"document_id": "doc_abc", "sections": [{"offset": 2000, "limit": 1000}]}]}

Read only the relevant section

uv run --script <skill-root>/commands/doc-read doc_abc --offset 2000 --limit 1000

undefined

uv run --script <skill-root>/commands/doc-read doc_abc --offset 2000 --limit 1000

undefined

Link Related Documents

关联相关文档

bash

undefined

bash

undefined

Get relation types

uv run --script <skill-root>/commands/doc-link --types

Create a sequential relationship

uv run --script <skill-root>/commands/doc-link --create doc_design doc_implementation --type predecessor-successor --from-to-note "Design doc" --to-from-note "Implementation plan"

Create hierarchical relationship (parent owns children)

uv run --script <skill-root>/commands/doc-link --create doc_architecture doc_api --type parent-child --from-to-note "API spec"

Create peer relationship

uv run --script <skill-root>/commands/doc-link --create doc_api doc_models --type related --from-to-note "Data models"

View document with its relations

uv run --script <skill-root>/commands/doc-info doc_architecture

---

uv run --script <skill-root>/commands/doc-info doc_architecture

---

Key Concepts

核心概念

Tag AND Logic

标签AND逻辑

IMPORTANT: Multiple tags means ALL must match:

```
--tags "python,api"
```
→ Document must have BOTH tags
```
--tags "v2,design,spec"
```
→ Document must have ALL THREE tags

重要提示：多个标签表示必须同时匹配所有标签：

```
--tags "python,api"
```
→ 文档必须同时包含这两个标签
```
--tags "v2,design,spec"
```
→ 文档必须同时包含这三个标签

Output Format

输出格式

All commands output JSON. Save document IDs from upload for later retrieval/deletion.

所有命令均输出JSON格式。请保存上传文档时返回的文档ID，以便后续检索或删除。

Document Relations

文档关联

Relation Types:

parent-child: Hierarchical ownership. Deleting parent cascades to children.
related: Peer relationship. No cascade delete.

Bidirectional: Relations are stored from both perspectives. Creating a parent-child link adds:

```
parent
```
relation on the parent document
```
child
```
relation on the child document

关联类型：

parent-child：层级归属关系。删除父文档会同时删除其子文档。
related：同级关联关系。删除文档不会影响关联的同级文档。

双向存储：关联关系会从两个方向存储。创建父子关联会同时添加：

父文档中的
```
parent
```
关联
子文档中的
```
child
```
关联

Notes for AI Assistants

AI助手注意事项

Tag AND logic - Multiple tags = ALL must match
Save document IDs - From upload output for future operations
Check server running - Handle connection errors gracefully
Parse JSON output - All commands return JSON
Tags are lowercase - Use consistent tag naming (
```
python
```
not
```
Python
```
)
Relations in doc-info - Use
```
doc-info
```
to see document relations without a separate call
Relation IDs - Save relation IDs from create output for update/remove operations
Cascade delete - Deleting a parent document also deletes its children

标签AND逻辑 - 多个标签表示必须同时匹配所有标签
保存文档ID - 保存上传输出中的文档ID以便后续操作
检查服务器状态 - 优雅处理连接错误
解析JSON输出 - 所有命令均返回JSON格式
标签小写 - 使用统一的小写标签命名（如
```
python
```
而非
```
Python
```
）
通过doc-info查看关联 - 使用
```
doc-info
```
命令查看文档关联，无需单独调用其他命令
保存关联ID - 保存创建关联时输出的关联ID以便后续更新或删除
级联删除 - 删除父文档会同时删除其子文档

Quick Decision Tree

快速决策树

Store document? →

doc-push <file> --tags "tag1,tag2"

Find by tags? →

doc-query --tags "tag1,tag2"

(AND logic, add

--include-relations

for links)

Find by meaning? →

doc-search "your question"

(add

--include-relations

for links)

Check metadata? →

doc-info <doc-id>

(metadata + relations)

Read text file? →

doc-read <doc-id>

(text files to stdout)

Read section? →

doc-read <doc-id> --offset 2000 --limit 1000

(partial read)

Download document? →

doc-pull <doc-id>

(ID from query)

Remove document? →

doc-delete <doc-id>

(permanent)

List all? →

doc-query

(no args)

Link documents? →

doc-link --create <from> <to> --type parent-child

List relation types? →

doc-link --types

Update link note? →

doc-link --update <id> --note "text"

Remove link? →

doc-link --remove <id>

(removes both directions)

需要存储文档？ →

doc-push <file> --tags "tag1,tag2"

需要按标签查找？ →

doc-query --tags "tag1,tag2"

（AND逻辑，添加

--include-relations

查看关联）

需要按语义查找？ →

doc-search "你的问题"

（添加

--include-relations

查看关联）

需要查看元数据？ →

doc-info <doc-id>

（包含元数据和关联信息）

需要读取文本文件？ →

doc-read <doc-id>

（将文本文件输出到标准输出）

需要读取特定段落？ →

doc-read <doc-id> --offset 2000 --limit 1000

（部分读取）

需要下载文档？ →

doc-pull <doc-id>

（文档ID来自查询结果）

需要删除文档？ →

doc-delete <doc-id>

（永久删除）

需要列出所有文档？ →

doc-query

（不带任何参数）

需要关联文档？ →

doc-link --create <from> <to> --type parent-child

需要查看关联类型？ →

doc-link --types

需要更新关联备注？ →

doc-link --update <id> --note "文本内容"

需要删除关联？ →

doc-link --remove <id>

（同时删除双向关联）

Additional Resources

额外资源

Detailed Command Reference: See
```
references/COMMANDS.md
```
Configuration Options: See
```
references/CONFIGURATION.md
```
Error Handling: See
```
references/TROUBLESHOOTING.md
```

详细命令参考：查看
```
references/COMMANDS.md
```
配置选项：查看
```
references/CONFIGURATION.md
```
错误处理：查看
```
references/TROUBLESHOOTING.md
```

context-store

Original

Translation

Context Store Skill

Context Store Skill

What & When

是什么与何时使用

Quick Reference

快速参考

doc-push - Upload Documents

doc-push - 上传文档

Example: uv run --script <skill-root>/commands/doc-push specs.md --tags "api,v2"

Example: uv run --script <skill-root>/commands/doc-push specs.md --tags "api,v2"

doc-query - Search by Tags/Name

doc-query - 按标签/名称搜索

Example: uv run --script <skill-root>/commands/doc-query --tags "api,v2"

Example: uv run --script <skill-root>/commands/doc-query --tags "api,v2"

With relations: uv run --script <skill-root>/commands/doc-query --tags "api" --include-relations

With relations: uv run --script <skill-root>/commands/doc-query --tags "api" --include-relations

doc-search - Semantic Search

doc-search - 语义搜索

Example: uv run --script <skill-root>/commands/doc-search "how to configure authentication"

Example: uv run --script <skill-root>/commands/doc-search "how to configure authentication"

With relations: uv run --script <skill-root>/commands/doc-search "auth" --include-relations

With relations: uv run --script <skill-root>/commands/doc-search "auth" --include-relations

doc-info - Get Document Metadata & Relations

doc-info - 获取文档元数据与关联信息

Example: uv run --script <skill-root>/commands/doc-info doc_abc123

Example: uv run --script <skill-root>/commands/doc-info doc_abc123

doc-read - Read Text Documents

doc-read - 读取文本文档

Example: uv run --script <skill-root>/commands/doc-read doc_abc123

Example: uv run --script <skill-root>/commands/doc-read doc_abc123

Partial: uv run --script <skill-root>/commands/doc-read doc_abc123 --offset 2000 --limit 1000

Partial: uv run --script <skill-root>/commands/doc-read doc_abc123 --offset 2000 --limit 1000

doc-pull - Download Documents

doc-pull - 下载文档

Example: uv run --script <skill-root>/commands/doc-pull doc_abc123 -o specs.md

Example: uv run --script <skill-root>/commands/doc-pull doc_abc123 -o specs.md

doc-delete - Remove Documents

doc-delete - 删除文档

Example: uv run --script <skill-root>/commands/doc-delete doc_abc123

Example: uv run --script <skill-root>/commands/doc-delete doc_abc123

doc-link - Manage Document Relations

doc-link - 管理文档关联

Typical Workflows

典型工作流

Store and Retrieve

存储与检索

Upload with tags

Upload with tags

Find it later

Find it later

Download it

Download it

Build Knowledge Base

构建知识库

Upload multiple documents with consistent tags

Upload multiple documents with consistent tags

Query by project phase

Query by project phase

Semantic Search + Partial Read

语义搜索+部分读取

Search by meaning

Search by meaning

Returns: {"results": [{"document_id": "doc_abc", "sections": [{"offset": 2000, "limit": 1000}]}]}

Returns: {"results": [{"document_id": "doc_abc", "sections": [{"offset": 2000, "limit": 1000}]}]}

Read only the relevant section

Read only the relevant section

Link Related Documents

关联相关文档

Get relation types

Get relation types

Create a sequential relationship

Create a sequential relationship

Create hierarchical relationship (parent owns children)

Create hierarchical relationship (parent owns children)

Create peer relationship

Create peer relationship

View document with its relations

`doc-push`
- Upload Documents

`doc-push`
- 上传文档

`doc-query`
- Search by Tags/Name

`doc-query`
- 按标签/名称搜索

`doc-search`
- Semantic Search

`doc-search`
- 语义搜索

`doc-info`
- Get Document Metadata & Relations

`doc-info`
- 获取文档元数据与关联信息

`doc-read`
- Read Text Documents

`doc-read`
- 读取文本文档

`doc-pull`
- Download Documents

`doc-pull`
- 下载文档

`doc-delete`
- Remove Documents

`doc-delete`
- 删除文档

`doc-link`
- Manage Document Relations

`doc-link`
- 管理文档关联